Known in the art are various devices for separating and cleaning gas flows from solid particles, e.g. from dust, such as: settling chambers, in which the largest solid particles (grit) settle by gravity; cyclones and inertial dust separators, which make use of centrifugal and inertial effects arising from changes in the direction of the gas flow; industrial filters (also known as "bag houses"), in which the dust-laden gas passes through cloth, layers of paper, glass wool, metal meshes, etc.; electrostatic precipitators (electro-filters) in which the particles are electrically charged in a high-voltage electric field and then drift to an electrode on which they settle; and other devices, such as wet scrubbers, in which the dust particles are brought in contact with a liquid and subsequently swept away.
When no particular system can provide the required degree of cleaning, equipment operating on more than one of the above approaches may be used (e.g., a cyclone separator may be combined with a fabric filter).
One of the main characteristics of a dust separator is its gravimetric cleaning efficiency .eta., which is commonly defined as the ratio of the weight of dust trapped to the weight of incoming dust (in the same period of time). The cleaning efficiency can be expressed either as a number .eta..ltoreq.1 (as, for the most part, will be done below) or, upon multiplication by 100, in percent.
A more detailed characterization of a dust-separating device is provided by the fractional efficiency, which is an indication of the variation of gravimetric efficiency as a function of the particle size d. The fractional efficiency may be expressed as a formula or a curve .eta.(d) which shows gravimetric efficiency in a continuous series of narrow ranges of the particle size spectrum (W. Strauss, "Industrial Gas Cleaning" Pergamon Press, 1966)
Generally speaking, the smaller the particles, the more difficult is the task of their separation. Correspondingly, the fractional efficiency curves .eta.(d) are not constant across the particle size range, but tend to rapidly fall off for smaller particle sizes, approaching zero efficiency as the particle size tends to zero. Thus, in characterizing various dust-separating devices, the question of importance is below which particle size the fractional efficiency starts to decrease appreciably. For example, for a typical gravity settling chamber, the fractional efficiency curve starts to decrease around the 80-100 .mu.m range, and may reach .eta.=0.8 (80%) efficiency around 50 .mu.m (Strauss, supra).
Taking, somewhat arbitrarily, a cleaning efficiency of 80% as the reasonable criterion dividing the more useful devices (.eta.&gt;80%) from the less useful ones (.eta.&lt;80%), it is found that only electro-filters, cloth filters, certain types of wet scrubbers, and special small-radius cyclones can separate particles less than 10 .mu.m in size with .eta..gtoreq.80% (Strauss, supra; "High-Efficiency Air Filtration", edited by P. A. F. White and S. E. Smith, Butterworths, London, 1964).
Another parameter which may affect the cleaning efficiency of a device is the dust concentration or dust density in the gas, measured in g/m.sup.3, at the input of the device.
Yet another important parameter associated with the operation of a dust-separating device and having an impact on its cleaning efficiency is the average gas velocity through the device. The existing dust-separating devices operate at different gas velocities, depending on the principle of their operation. For example, while electro-filters operate at a relatively low gas velocity seldom exceeding 2 m/s, inertial dust separators work at speeds between 15 and 30 m/s. The higher velocities are generally desirable, since they imply higher throughput of a device (the latter being the product of a gas flow velocity and the cross-sectional area of the gas flow through the device). For a given required throughput, the higher the allowed gas velocity, the smaller can be the size of the device.
However, as the gas velocity exceeds a certain optimum value (which depends on the type of device being used), the cleaning efficiency starts to drop, sometimes abruptly. For example, in inertial separators this drop in the cleaning efficiency occurs due to a set up of strong turbulence in a gas flow at speeds over 30 m/s. At the same time, by virtue of their nature, the inertial separators also lose their cleaning efficiency at low gas velocities. In real devices of this type, the useful range of gas velocities is typically rather narrow, e.g. .+-.20% of the optimum velocity value at which .eta. is maximum.
When two or more (in general, n) devices, which individually have cleaning efficiencies .eta..sub.1, .eta..sub.2, . . . , .eta..sub.n, are connected in series, so that the cleaned gas from an upstream device enters the next downstream device, it is straightforward to show that the total cleaning efficiency of n devices in series can be expressed as EQU E=.eta..sub.1 +(1-.eta..sub.1).eta..sub.2 +(1-.eta..sub.1) (1-.eta..sub.2).eta..sub.3 +. . . +(1-.eta..sub.1)(1-.eta..sub.2) . . . (1-.eta..sub.n-1) .eta..sub.n ( 1)
with .eta. being expressed as a number smaller than unity rather than in percent. For example, for two separators with individual efficiencies .eta..sub.1 and .eta..sub.2 connected in series, EQU E=.eta..sub.1 +(1-.eta..sub.1).eta..sub.2. (2)
In light of what has been said about the fractional efficiency .eta.(d) and the dust density, it will be apparent that formulas (1) and (2) should be used with caution. Indeed, the first device of a series will tend to separate out primarily the larger particles and to supply the next downstream device with the partially cleaned gas which will have both (a) a lower dust density and (b) a particle size distribution with, generally, smaller average particle size as compared to the size distribution of dust entering the first device. This shift in particle size distribution is a direct consequence of the non-constant fractional efficiency curve .eta.(d); it places a practical limit to the degree of particle separation from the gas flow which can be achieved by means of plural cleaning devices connected in series. Thus, the values of .eta. in formula (1) and (2) should be understood as corresponding to the characteristics of a dust-carrying gas flow at the inputs of each respective dust separating device. In practice these values are obtained experimentally.
In the range of particle sizes in which .eta.(d) does not become too small, connecting a plurality of devices in series is an effective way to increase the total cleaning efficiency of a system. For example, from formula (2), taking .eta..sub.1 =0.7 and .eta..sub.2 =0.6, it is found that E=0.88 (88%).
Clearly, devices with an essentially flat .eta.(d) down to small particle sizes would be especially useful for high purification of gas flows by means of connecting them in series.
As those skilled in the field of multiple-component fluid (i.e., gas/solids) separation know, the prior art includes the above-mentioned class of devices known as inertial or momentum separators, in which the the gas is cleaned of solid particles by utilizing abrupt changes in the direction of movement of the gas flow and a reduction in its velocity. The solid particles, because of their inertia, will continue to move in the same direction as the initial gas flow, and will eventually be deposited into a collecting hopper. The heavier (larger) particles have more inertia and thus are cleaned with more ease than the lighter (smaller) ones, which tend to escape with the cleaned gas flow.
Some of these devices are constructed so as to provide a number of solid (typically, metal) surfaces inclined at an acute angle with respect to the gas flow, the aim of the surfaces being to deflect the solid particles away from their paths which originally coincide with the direction of the main gas flow. The surfaces thus help to concentrate the solid particles on the one side of the said deflecting parts, while the cleaned gas escapes through the spaces between the deflecting parts.
Such a device is the shutter-type collector (Strauss, supra; C. J. Stairmand, Trans. Inst. Chem. Eng. (London), 29, 356 (1951)) which is sometimes used as a pre-cleaning stage before cyclones or bag houses. More efficient is the conical louvered collector and its variations (Strauss, supra; K. Hansen, Fifth World Power Conference, Vienna, 16, 5829 (1956); E. Haber, U.S. Pat. No. 2,034,467 and U.K. Pat. No. 388,637; H. Van Der Kolk, U.S. Pat. No. 2,874,800 and U.K. Pat. No. 766,279; I. R. W. Johnston, U.S. Pat. No. 4,340,474; H. Keller, U.S. Pats. No. 3,958,966 and No. 4,198,220; K.H. Maden, U.S. Pat. No. 4,123,241). In a popular version (see, for example, the Haber patents), it consists of a system of conically mounted flat-surfaced conical rings of decreasing diameters. The flat-surfaced conical rings are mounted in a cylindrical or conical casing so that they overlap each other in the axial direction leaving narrow gaps between adjacent ring surfaces. These annular gaps are oriented at a sharp angle with respect to the direction of the gas flow. Gas flow is supplied to and enters the casing at the end thereof adjacent the location of the ring of the largest diameter, and moves through the cone from the top down. The main part of the gas containing the lighter particles abruptly changes the direction of its motion and escapes upwardly through the inter-ring gaps for discharge to its next destination, while the larger particles continue to move downwardly through the cone. At the same time, the particles repeatedly impact on the ring surfaces (which incidentally, leads to considerable ring wear over a period of time) and are thereby projected towards the axis of the cone; they are thus concentrated and are removed with a part of the gas (typically 5-7% ) through the ring of the smallest diameter.
In a variation of this device (see the Van Der Kolk patents), a conical one-piece construction is made of a spirally wound wire having an either rectangular or trapezoidal cross-section, with the straight inside wire surface being obliquely inclined with respect to the cone axis and serving the same purpose as the flat conical ring in the previously described device.
The advantages of the above-described known conical inertial dust collectors reside in the simplicity and compactness of their design, an absence of moving parts, a relatively low drag for the gas flow, and a relatively high gas flow velocity through the device (i.e., a high throughput), as well as in that their efficiency does not change much with variations in the input dust density (Strauss, supra; Hansen, supra). The disadvantages of these devices are their inability to effectively remove particles smaller than 20-30 .mu.m, and a relatively large amount of gas which does not get separated from the solid particles and has to be subsequently cleaned with downstream cyclones (Strauss, supra). That is in essence why conical inertial dust traps are mostly used as pre-cleaning devices, for removal of coarser particles.
Another disadvantage of the conical inertial dust collectors is the constant bombardment of the conical rings by solid particles, which in some cases leads to their relatively rapid erosion and wear, thus necessitating frequent maintenance, including ring replacement, and consequent system down-time.
Inertial dust separators with more elaborate, curvilinear particle-deflecting elements are also known. The Johnston patent, for example, describes a conical-type device with axially non-overlapping rings of a more complex shape, inwardly curved on the inside, and having a straight section and a lip designed in order to deflect solid particles toward the axis of the cone. The claimed total cleaning efficiency for the particles with average particle size between 20 and 30 .mu.m is close to 80%. However, no detailed fractional efficiency data are given.
In another device (see the Keller patents), the gas flow is directed against the tip, i.e., from the narrowest to the widest end, of a conically shaped separator (which is thus inverted as compared to the cone-shaped inertial separator described above), with individual separator elements or rings overlapping each other and having their cross-section shaped as an obtuse triangle. The idea here is to provide deflection surfaces for the solid particles, just as before, although in an inverted geometry, whereby the concentrated particles are moving outside of a cone structure toward the wider base of a cone, while the cleaned gas escapes into the inside volume of the cone.
By virtue of the complex ring shape, curved channels are formed between the rings, the channels serving to provide escape passages for the partially cleaned gas flow. The passages thus formed open up toward the inside of a cone structure. This feature is supposed to help to prevent clogging of a device. Again, however, no experimental data are given. Moreover, it should also be noted that the device is intended primarily for separating solid particles from steam, which may be the reason for special attention paid to the clogging problem.
A somewhat similar device is described in the Maden patent in which one or two sets of particle deflectors or "vanes" are positioned in a hollow body of rectangular, circular or elongated shape as viewed in a cross-section normal to the flow direction. As in the Keller patents, in cross-section the vanes have a complex, curved, elongated triangular shape. They are positioned in the body so as to provide deflection surfaces for the incoming solid particles (with the preferred angle of the particle-deflecting front surface being at 34.degree. to the axis of the device), and also so as to mask the passages between them from the direct hit by an incoming particle. By virtue of their inertia, solid particles (at least the larger ones) miss the passages between the vanes, and continue to move down the narrowing device, while the cleaned gas escapes between the vanes, through the curved passages defined in each case by the overlapping trailing surface of an upstream vane and the long back surface of the next adjacent downstream vane. Due to the concave shape of the trailing surface of each vane, the passage opens up toward the outside space between the system of vanes and the body of the device, as was also the case in the Keller patents. In addition, the passages between adjacent vanes may have a changing shape, with increasingly tortuous passages toward the outlet end of the device. This device is claimed to produce a cleaning efficiency of up to 89.7%, although it is not specified in the patent what kind of dust has been used for the test, nor even the average particle size. About 10% of the gas is lost through the outlet of the device together with concentrated dust.